Columnar: add explain info for chunk filtering (#4554)

Co-authored-by: Jeff Davis <jefdavi@microsoft.com>

src/backend/columnar/cstore_customscan.c

@@ -68,6 +68,8 @@ static void CStoreScan_BeginCustomScan(CustomScanState *node, EState *estate, in
 static TupleTableSlot * CStoreScan_ExecCustomScan(CustomScanState *node);
 static void CStoreScan_EndCustomScan(CustomScanState *node);
 static void CStoreScan_ReScanCustomScan(CustomScanState *node);
+static void CStoreScan_ExplainCustomScan(CustomScanState *node, List *ancestors,
+										 ExplainState *es);
 
 /* saved hook value in case of unload */
 static set_rel_pathlist_hook_type PreviousSetRelPathlistHook = NULL;
@@ -93,7 +95,7 @@ const struct CustomExecMethods CStoreExecuteMethods = {
 	.EndCustomScan = CStoreScan_EndCustomScan,
 	.ReScanCustomScan = CStoreScan_ReScanCustomScan,
 
-	.ExplainCustomScan = NULL,
+	.ExplainCustomScan = CStoreScan_ExplainCustomScan,
 };
 
 
@@ -439,4 +441,19 @@ CStoreScan_ReScanCustomScan(CustomScanState *node)
 }
 
 
+static void
+CStoreScan_ExplainCustomScan(CustomScanState *node, List *ancestors,
+							 ExplainState *es)
+{
+	TableScanDesc scanDesc = node->ss.ss_currentScanDesc;
+
+	if (scanDesc != NULL)
+	{
+		int64 chunksFiltered = ColumnarGetChunksFiltered(scanDesc);
+		ExplainPropertyInteger("Columnar Chunks Removed by Filter", NULL,
+							   chunksFiltered, es);
+	}
+}
+
+
 #endif /* HAS_TABLEAM */

src/backend/columnar/cstore_reader.c

@@ -44,7 +44,8 @@ static StripeBuffers * LoadFilteredStripeBuffers(Relation relation,
 												 StripeMetadata *stripeMetadata,
 												 TupleDesc tupleDescriptor,
 												 List *projectedColumnList,
-												 List *whereClauseList);
+												 List *whereClauseList,
+												 int64 *chunksFiltered);
 static void ReadStripeNextRow(StripeBuffers *stripeBuffers, List *projectedColumnList,
 							  uint64 chunkIndex, uint64 chunkRowIndex,
 							  ChunkData *chunkData, Datum *columnValues,
@@ -54,7 +55,8 @@ static ColumnBuffers * LoadColumnBuffers(Relation relation,
 										 uint32 chunkCount, uint64 stripeOffset,
 										 Form_pg_attribute attributeForm);
 static bool * SelectedChunkMask(StripeSkipList *stripeSkipList,
-								List *projectedColumnList, List *whereClauseList);
+								List *projectedColumnList, List *whereClauseList,
+								int64 *chunksFiltered);
 static List * BuildRestrictInfoList(List *whereClauseList);
 static Node * BuildBaseConstraint(Var *variable);
 static OpExpr * MakeOpExpression(Var *variable, int16 strategyNumber);
@@ -104,6 +106,7 @@ CStoreBeginRead(Relation relation, TupleDesc tupleDescriptor,
 	readState->stripeBuffers = NULL;
 	readState->readStripeCount = 0;
 	readState->stripeReadRowCount = 0;
+	readState->chunksFiltered = 0;
 	readState->tupleDescriptor = tupleDescriptor;
 	readState->stripeReadContext = stripeReadContext;
 	readState->chunkData = NULL;
@@ -153,7 +156,9 @@ CStoreReadNextRow(TableReadState *readState, Datum *columnValues, bool *columnNu
 																 readState->
 																 projectedColumnList,
 																 readState->
-																 whereClauseList);
+																 whereClauseList,
+																 &readState->
+																 chunksFiltered);
 		readState->readStripeCount++;
 		readState->currentStripeMetadata = stripeMetadata;
 
@@ -332,7 +337,7 @@ CStoreTableRowCount(Relation relation)
 static StripeBuffers *
 LoadFilteredStripeBuffers(Relation relation, StripeMetadata *stripeMetadata,
 						  TupleDesc tupleDescriptor, List *projectedColumnList,
-						  List *whereClauseList)
+						  List *whereClauseList, int64 *chunksFiltered)
 {
 	uint32 columnIndex = 0;
 	uint32 columnCount = tupleDescriptor->natts;
@@ -345,7 +350,7 @@ LoadFilteredStripeBuffers(Relation relation, StripeMetadata *stripeMetadata,
 														stripeMetadata->chunkCount);
 
 	bool *selectedChunkMask = SelectedChunkMask(stripeSkipList, projectedColumnList,
-												whereClauseList);
+												whereClauseList, chunksFiltered);
 
 	StripeSkipList *selectedChunkSkipList =
 		SelectedChunkSkipList(stripeSkipList, projectedColumnMask,
@@ -474,7 +479,7 @@ LoadColumnBuffers(Relation relation, ColumnChunkSkipNode *chunkSkipNodeArray,
  */
 static bool *
 SelectedChunkMask(StripeSkipList *stripeSkipList, List *projectedColumnList,
-				  List *whereClauseList)
+				  List *whereClauseList, int64 *chunksFiltered)
 {
 	ListCell *columnCell = NULL;
 	uint32 chunkIndex = 0;
@@ -527,6 +532,7 @@ SelectedChunkMask(StripeSkipList *stripeSkipList, List *projectedColumnList,
 			if (predicateRefuted)
 			{
 				selectedChunkMask[chunkIndex] = false;
+				*chunksFiltered += 1;
 			}
 		}
 	}

src/backend/columnar/cstore_tableam.c

@@ -1156,6 +1156,26 @@ cstore_tableam_finish()
 }
 
 
+/*
+ * Get the number of chunks filtered out during the given scan.
+ */
+int64
+ColumnarGetChunksFiltered(TableScanDesc scanDesc)
+{
+	CStoreScanDesc cstoreScanDesc = (CStoreScanDesc) scanDesc;
+	TableReadState *readState = cstoreScanDesc->cs_readState;
+
+	if (readState != NULL)
+	{
+		return readState->chunksFiltered;
+	}
+	else
+	{
+		return 0;
+	}
+}
+
+
 /*
  * Implementation of TupleTableSlotOps.copy_heap_tuple for TTSOpsColumnar.
  */

src/include/columnar/cstore.h

@@ -236,6 +236,7 @@ typedef struct TableReadState
 	StripeBuffers *stripeBuffers;
 	uint32 readStripeCount;
 	uint64 stripeReadRowCount;
+	int64 chunksFiltered;
 	ChunkData *chunkData;
 	int32 deserializedChunkIndex;
 } TableReadState;

src/include/columnar/cstore_tableam.h

@@ -18,7 +18,7 @@ extern TableScanDesc cstore_beginscan_extended(Relation relation, Snapshot snaps
 											   ParallelTableScanDesc parallel_scan,
 											   uint32 flags, Bitmapset *attr_needed,
 											   List *scanQual);
-
+extern int64 ColumnarGetChunksFiltered(TableScanDesc scanDesc);
 extern bool IsCStoreTableAmTable(Oid relationId);
 extern TableDDLCommand * ColumnarGetTableOptionsDDL(Oid relationId);
 extern char * GetShardedTableDDLCommandColumnar(uint64 shardId, void *context);

src/test/regress/input/am_chunk_filtering.source

@@ -71,3 +71,9 @@ B
 \.
 
 SELECT * FROM collation_chunk_filtering_test WHERE A > 'B';
+
+CREATE TABLE simple_chunk_filtering(i int) USING COLUMNAR;
+INSERT INTO simple_chunk_filtering SELECT generate_series(0,234567);
+EXPLAIN (analyze on, costs off, timing off, summary off)
+  SELECT * FROM simple_chunk_filtering WHERE i > 123456;
+DROP TABLE simple_chunk_filtering;

src/test/regress/output/am_chunk_filtering.source

@@ -118,3 +118,16 @@ SELECT * FROM collation_chunk_filtering_test WHERE A > 'B';
  Å
 (1 row)
 
+CREATE TABLE simple_chunk_filtering(i int) USING COLUMNAR;
+INSERT INTO simple_chunk_filtering SELECT generate_series(0,234567);
+EXPLAIN (analyze on, costs off, timing off, summary off)
+  SELECT * FROM simple_chunk_filtering WHERE i > 123456;
+                                    QUERY PLAN
+---------------------------------------------------------------------
+ Custom Scan (ColumnarScan) on simple_chunk_filtering (actual rows=111111 loops=1)
+   Filter: (i > 123456)
+   Rows Removed by Filter: 3457
+   Columnar Chunks Removed by Filter: 12
+(4 rows)
+
+DROP TABLE simple_chunk_filtering;